Bromodomain and Extra-terminal (BET) Protein Inhibitors Suppress Chondrocyte Differentiation and Restrain Bone Growth*

Small molecule inhibitors for bromodomain and extra-terminal (BET) proteins have recently emerged as potential therapeutic agents in clinical trials for various cancers. However, to date, it is unknown whether these inhibitors have side effects on bone structures. Here, we report that inhibition of BET bromodomain proteins may suppress chondrocyte differentiation and restrain bone growth. We generated a luciferase reporter system using the chondrogenic cell line ATDC5 in which the luciferase gene was driven by the promoter of Col2a1, an elementary collagen of the chondrocyte. The Col2a1-luciferase ATDC5 system was used for rapidly screening both activators and repressors of human collagen Col2a1 gene expression, and we found that BET bromodomain inhibitors reduce the Col2a1-luciferase. Consistent with the luciferase assay, BET inhibitors decrease the expression of Col2a1. Furthermore, we constructed a zebrafish line in which the enhanced green fluorescent protein (EGFP) expression was driven by col2a1 promoter. The transgenic (col2a1-EGFP) zebrafish line demonstrated that BET inhibitors I-BET151 and (+)-JQ1 may affect EGFP expression in zebrafish. Furthermore, we found that I-BET151 and (+)-JQ1 may affect chondrocyte differentiation in vitro and inhibit zebrafish growth in vivo. Mechanistic analysis revealed that BET inhibitors influenced the depletion of RNA polymerase II from the Col2a1 promoter. Collectively, these results suggest that BET bromodomain inhibition may have side effects on skeletal bone structures.

BET proteins and BET inhibitors have been reported to be involved in a variety of biological processes. Pharmacological inhibition of BET proteins lowers the expression of key transcription factors such as oncogene c-MYC (8), clamps the transductions of PI3K signaling (9), inhibits Gli1 transcription and Hedgehog pathway (10), blocks transcription in neurons (11), represses VEGF-induced angiogenesis and vascular permeability (12), reduces cell viability of osteosarcoma cells and inhibits osteoblastic differentiation (13), and restrains osteoclastogenesis (14). Meanwhile, potent BET inhibitors have been identified as showing antitumor efficacy in a number of preclinical cancer models in recent years, including leukemia, multiple myeloma, lymphoma, melanoma, and gastric cancer (15,16). This led to clinical studies focusing mostly on the treatment of leukemia and lymphoma. As a result of these studies, the first encouraging signs of efficacy have already been reported (17). Furthermore, previous studies showed that BET inhibitors also control neuronal differentiation and cause an autism-like syndrome (18). However, BET inhibitors were not examined extensively in these studies to determine their side effects on skeletal bone structures.
The long bone is mainly formed through endochondral bone formation, which starts with the formation of a cartilage template from condensed mesenchymal cells. The chondrocytes of the cartilage template proliferate axially and subsequently undergo hypertrophy and expansion in cellular volume (19). In addition to cartilage development, reports show that the chondrocytes secrete a variety of collagen proteins (20). Type II and X collagens (Col2a1 and Col10a1) contribute to chondrocyte maintenance and proliferation (20,21). Hypertrophic chondro-cytes will then switch from the synthesis of the cartilage-characteristic collagens to the synthesis of type I collagen (Col1a1), an osteoblast-characteristic collagen that organizes a mineralizing bone matrix (20,21).
To further examine the underlying mechanisms for chondrocyte proliferation and growth inhibition, we established a chondrogenic ATDC5 cell line with the Col2a1 promoter (Col2a1-luc ATDC5 system) and screened a list of epigenetic compounds. Strikingly, six different BET inhibitors were shown to reduce Col2a1-luc, consistent with the decrease of Col2a1 mRNA and protein. Furthermore, we found that BET inhibitors blocked the differentiation of chondrocyte culture. Consistent with this, BET inhibitors I-BET151 and (ϩ)-JQ1 could induce retardation of the growth of zebrafish. Taken together, these data suggest that the Col2a1-luc ATDC5 system may be used to screen chondrogenic factors. Furthermore, these data suggest that anticarcinogenic BET inhibitors may potentially injure chondrocytes and the bones of patients.

Results
BET Inhibitors Suppressed the Expression of Col2a1 in Chondrogenic ATDC5 Cells-Type II collagen is the specific and major matrix protein of cartilage. The Col2a1 promoter has been used to identify the molecules related to chondrogenesis (22,23). To explore novel activators or repressors of chondrogenic differentiation, we generated an ATDC5 cell line that includes the 3-kb human Col2a1 promoter ligated to the open reading frame of firefly luciferase (Fig. 1A). We checked the responses of Col2a1-luc to the heterocyclic molecule kartogenin (KGN) and transcriptional factor SOX9, which have been widely recognized to promote the expression of Col2a1 and chondrocyte differentiation (24,25). In agreement with previous reports (24,25), both KGN and SOX9 increased Col2a1 promoter activities in a dose-dependent manner (Fig. 1, B and C), suggesting that the Col2a1-luc ATDC5 system could be used to screen chondrogenic activators or repressors.
Epigenetic factors (i.e. chromatin modifiers) regulate normal and disease processes and are mediated in part by the methylation status of DNA as well as by chemical modifications of histones, including acetylation, methylation, phosphorylation, and ubiquitination (26). Chondrogenic processes are also influenced by epigenetic regulation, such as methylation of Lys-27 residues on histone H3 (27). To uncover more epigenetic regulators, we screened a collection of 38 chemicals involved in epigenetic regulations using the Col2a1-luc ATDC5 system (Table 1). After 48-h treatment in 10 M final concentration of BET inhibitors, we found that a battery of epigenetic inhibitors decreased luciferase activities of Col2a1 promoter ( Fig. 2A). Among them, all BET inhibitors, including bromosporine, I-BET-762, RVX208, I-BET151, PFI-1, and (ϩ)-JQ1, could significantly suppress the luciferase activity of the Col2a1 promoter ( Fig. 2A) compared with the non-response of the promoter of a soybean translation elongation factor EF1␣ (Fig. 2B).
Consistent with the luciferase activity result, the RNA levels of Col2a1 were also significantly reduced in cells treated with BET inhibitors compared with vehicle alone (DMSO) (Fig. 2C). Consistently, we discovered that BET inhibitors decreased Col2a1 expression in mesenchymal progenitor cell line C3H/ 10T1/2 (C3H10) (Fig. 2, D-F) and primary chondrocytes (Fig. 2, The half-maximal inhibitory concentration and cytotoxicity concentration were unequal for different BET inhibitors. To lower the impact of cell cytotoxicity of the various BET inhibitors, we further examined the effects of five BET inhibitors (I-BET-762, RVX208, I-BET151, PFI-1, and (ϩ)-JQ1) on Col2a1 expression at various doses. We performed treatment with five BET inhibitors at gradient concentrations of 0.1, 1, and 5 M. As shown in Fig. 3, A-E (ATDC5) and F-J (C3H10), all of the BET inhibitors were capable of dose-dependently decreasing the luciferase activity of the Col2a1 promoter. Consistent with this, the expression levels of Col2a1, as determined by quantitative RT-PCR, were dose-dependently reduced by various BET inhibitors (Fig. 3, K-O (ATDC5) and P-T (primary chondrocytes)).
BET Inhibitors Dose-dependently Suppressed the Expression of Chondrogenic Marker Genes in Primary Chondrocytes-BET bromodomain inhibition results in the suppression of Col2a1, which is one of the most essential collagens expressed during cartilage maintenance and differentiation. We hypothesized that BET inhibitors may also play fundamental roles in chondrocyte proliferation and development. To test our hypothesis, we isolated chondrocytes from the articular cartilage of 3-dayold wild-type mice and examined the effects of BET bromodomain inhibition on the expression of chondrocyte marker genes. Total RNA and protein were collected from the cultured chondrocytes that were treated with I-BET151 and (ϩ)-JQ1 at final concentrations of 0.2, 1, and 5 M (I-BET151) and 0.2, 0.5, and 1 M (ϩ)-JQ1) for 72 h. Using quantitative RT-PCR, we observed a progressive transcription suppression of Col2a1, a disintegrin and metalloproteinase with thrombospondin motifs 5 (Adamts5), and matrix metalloproteinase-3 (Mmp3) ( Fig. 4, A-F). However, the key transcriptional factor of chondrogenesis, Sox9, showed no obvious response to the BET bromodomain inhibition (data not shown). We also examined the protein levels of different chondrocyte marker genes after I-BET151 and (ϩ)-JQ1 treatment. As shown in Fig. 4, G and H, immunoblotting analysis demonstrated that both BET151 and (ϩ)-JQ1 may lead to decreased protein levels of COL2A1, ADAMTS5, and MMP3, correlating with RNA level inhibition. These data indicated that both I-BET151 and (ϩ)-JQ1 treatments may have suppressive effects on chondrogenesis.
BET Inhibition with I-BET151 and (ϩ)-JQ1 Arrests Chondrogenesis and Differentiation-To further characterize the efficacy of inhibiting BET proteins in chondrogenic differentiation, we examined the effect of I-BET151 and (ϩ)-JQ1 by using the chondrocyte micromass culture system. Inhibition of BET bromodomain proteins induced a remarkable decrease in matrix accumulation, which was monitored by Alcian blue staining after 7 days in micromass culture (Fig. 5A). Accordingly, both I-BET151 and (ϩ)-JQ1 strongly suppressed the expression of chondrogenesis-related genes, such as Col2a1, Adamts5, and Mmp3 (Fig. 5, B-D). Consistently, the chondrogenic differentiation of C3H10 and ATDC5 cells was inhibited by I-BET151 and (ϩ)-JQ1 as well (Fig. 5, E and F).
Next, we examined the effects of BET inhibitors on ex vivo mouse articular cartilage cultures. Articular cartilage of distal femora was isolated from newborn mice and cultured in chon-drogenic differential medium. I-BET151 at 5 M or (ϩ)-JQ1 at 0.5 M was added to the culture medium with an equal amount of DMSO as the control. After 1 week of culture, we observed that both I-BET151 and JQ1 could suppress the growth of cartilage as the size of the I-BET151 and JQ1 treatment group was significantly smaller than the DMSO control group (Fig. 5, G-J). We examined the cartilage by positive safranin O staining and found that both I-BET151 and (ϩ)-JQ1 can inhibit the hypertrophy of chondrocytes as determined by the size of the chondrocytes following treatment (Fig. 5

, G and H). I-BET151 and (ϩ)-JQ1 Modulated the Accumulation of Pol II on Col2a1
Promoter-BET inhibitors are capable of blocking the interaction of acetylated lysine with BET proteins, which could regulate gene transcription driven by Pol II (28). First, we detected that the protein level of Pol II was not impacted by BET inhibitor treatment (Fig. 6A). We next examined whether BET inhibitors could affect the recruitment of Pol II to the Col2a1 promoter. For the 2.5-kb promoter upstream of the transcription start site of mouse Col2a1, we analyzed four loci (Fig. 6B, primers 1-4). DNA was purified after chromatin immunoprecipitation with Pol II antibody in primary chondrocytes treated with DMSO, I-BET151, or (ϩ)-JQ1 for 72 h. The relative amount of promoter-associated Pol II was evaluated by quantitative RT-PCR followed by normalization with the IgG sample. We observed that Pol II associates with the Col2a1 promoter, especially DNA fragments close to the transcription start site (Fig. 6C). In addition, the association between Pol II and the Col2a1 promoter was arrested by BET bromodomain inhibition by I-BET151 and (ϩ)-JQ1 (Fig. 6C). These data indicate that the recruitment of Pol II to the Col2a1 promoter was interrupted by BET inhibitors. I-BET151 and (ϩ)-JQ1 Retarded the Growth of Zebrafish-Zebrafish is a good, widely used model in developmental and functional studies of bone and cartilage (29). To determine whether BET inhibitors could influence Col2a1 as well as animal growth in vivo, we next examined the effects of BET inhibitors on zebrafish. We generated a transgenic zebrafish line harboring a col2a1 promoter-EGFP transgene (Fig. 7A). As shown in Fig. 7B, the EGFP fluorescence could be detected at 1 day postfertilization (dpf ) and was further enhanced at 2 dpf. The  (Fig. 7, C and D).
Prior research studies have demonstrated that long bones are formed through endochondral ossification of chondrocytes in which col2a1 is one of the most requisite matrix collagens (30). Thus, we hypothesized that the inhibition of BET bromodomain proteins may also affect the growth of the organism. To test this hypothesis, we incubated 2-dpf embryos with I-BET151 (1 and 5 M) and (ϩ)-JQ1 (0.1 and 0.5 M) for 2 weeks and measured the body length of the zebrafish following 2 weeks of cultivation. As shown in Fig. 7, E and F, zebrafish treated with I-BET151 at both 1 and 5 M developed a specified amount of atrophy compared with vehicle alone (DMSO). However, I-BET151 at high concentrations (10 M) impacted zebrafish viability (Fig. 7G). Consistently, we also observed that 0.5 M (ϩ)-JQ1 suppressed the length extension of the zebrafish, whereas 0.1 M (ϩ)-JQ1 showed no obvious functions (Fig. 7, H-J). Next, we characterized the whole mount skeleton of zebrafish by alizarin red and Alcian blue staining at 3 weeks old and observed similar shortened body length (Fig. 7,  K and L). To analyze the bone growth restriction, we measured the length of centrum 15 of caudal vertebrae as a representative and observed the shorter vertebral bodies in I-BET151 (both 1 and 5 M)and (ϩ)-JQ1 (0.5 M)-treated zebrafish, whereas 0.1 M (ϩ)-JQ1 showed no obvious functions (Fig. 7, M-P).
To demonstrate that BET inhibitors also played fundamental roles in zebrafish chondrogenesis, we isolated total RNAs of tail fins from 3-week-old zebrafish under different treatments. Using quantitative RT-PCR, we observed a progressive transcription suppression of cartilage and bone marker genes, such as col2a1, adamts5, col1a1 (type I collagen), and runt-related transcription factor 2a and 2b (runx2a and runx2b) (Fig. 8,  A-J). Consistent with this, the protein level of COL2A1 was  DECEMBER 23, 2016 • VOLUME 291 • NUMBER 52 JOURNAL OF BIOLOGICAL CHEMISTRY 26651 reduced in paraffin sections of 3-week-old zebrafish treated with both I-BET151 and (ϩ)-JQ1 (Fig. 8K). Taken together, our data suggest that BET inhibitors restrained chondrogenesis and differentiation both in vitro and in vivo.

Discussion
In the present study, we generated a Col2a1-luc ATDC5 system to identify activators or repressors of chondrogenic differentiation. We found that six BET inhibitors restrain the expression of Col2a1 and impede chondrogenesis as determined by micromass culture and ex vivo cartilage culture. We also established a Col2-EGFP transgenic zebrafish line and demonstrated that BET inhibitors can decrease Col2-EGFP expression. Consistently, BET inhibitors retard the growth of zebrafish in vivo. These data demonstrate that BET inhibitors could inhibit chondrogenic differentiation and restrain bone growth in vitro and in vivo.
The BET subfamily of bromodomain-containing proteins is involved in a number of hematological and solid tumors (15,17). Currently, several active clinical trials aiming to treat malignancies are using BET inhibitors, such as FT-1101, ZEN003694, BMS-986158, INCB054329, RVX-208, I-BET 762, OTX 015, CPI-0610, and TEN-010 (www.clinicaltrials.gov). In addition to this, multiple other BET inhibitors have been well studied and have revealed great potential for clinical application. For instance, (ϩ)-JQ1 suppresses breast cancer cells (31), gastric cancer cells (32), acute myeloid leukemia cells (33), and osteosarcoma cells (13). Furthermore, I-BET151 has been shown to induce cell apoptosis in myeloma and leukemia (16,34). However, because BET bromodomain proteins function as epigenetic regulators and play a key role in transcriptional regulation of essential genes in the cell cycle and apoptosis, such as c-Myc (8), nuclear factor -light chain enhancer of activated B cells (NF-B) (35), and B-cell lymphoma 2 (BCL2) (8,36), the inhibitory factors may affect other important biological processes.
Our current study demonstrated that BET inhibitors could interfere with chondrogenic differentiation and bone growth, highlighting the necessity of studying the role of bromodomain-containing proteins in chondrogenesis to monitor and avoid possible side effects of BET inhibitors on cartilage and bone. Besides bone formation, bone homeostasis is maintained by the balance between osteoblastic bone formation and osteoclastic bone resorption, whereas the growth of long bones is from endochondral ossification of chondrocytes derived from cartilage (15). Previous studies have shown that BET inhibitors may affect osteoclastogenesis (14) and the bone-associated tumor vicious cycle (13). In this study, we demonstrated BET inhibitors as repressors of chondrogenesis. Taken together, previous studies in addition to our studies demonstrate that preclinical pharmacological studies of BET inhibitors on human bone warrant further investigation.
Phenotype Analyses-For histology analyses, cartilage-like particles were fixed in 4% paraformaldehyde overnight at 4°C, dehydrated using graded ethanol overnight, and embedded in paraffin. The paraffin-embedded tissue samples were sectioned at 5 m. For safranin O staining, sections were first dewaxed, immersed in 1% acetic acid, and stained with 0.05% fast green and 0.5% safranine. Immunohistochemistry staining was performed as described previously (37). Images were captured using a microscope (Olympus BX51). Alizarin red and Alcian blue staining was performed as described previously (38).
Zebrafish were maintained as described (39). For compound treatment, DMSO and BET inhibitors were added as indicated at 1 dpf. Images were captured using an Olympus BZX16 stereomicroscope.
Luciferase Reporter Assays-ATDC5 cells grown in 10-cm dishes were transiently transfected using polyethylenimine (765090, Sigma-Aldrich) with the luciferase reporter constructs. pRL-TK (Promega, Madison, WI) was co-transfected to provide a normalization control for transfection efficiency. Eight hours after transfection, ATDC5 cells with luciferase and  DECEMBER 23, 2016 • VOLUME 291 • NUMBER 52

BET Inhibitors Induce Cartilage Destruction
control Renilla were digested and plated at 1.5 ϫ 10 5 /well in 12-well plate. Four hours later, cells were treated with different compounds. I-BET151 and (ϩ)-JQ1 were first dissolved in DMSO at 5 mM and then diluted to final concentrations as described. The vehicle, DMSO, was added as a blank control. After 48-h cultivation, cells were lysed, and luciferase activity was measured using the Dual-Luciferase Assay kit (Promega).
Statistical Analysis-For statistical analysis, in most graphs mean values with standard deviation (S.D.) are shown. Data were generated from several independently obtained data sets. p values were obtained from two-tailed student t tests. A p value less than 0.05 was considered a significant difference, and a p value less than 0.01 was considered an extremely significant difference. Error bars represent S.D.
Author Contributions-W. Z. and N. N. conceived and supervised the project. R. S. and N. N. generated the Col2a1-luc ATDC5 system and performed the screening of the epigenetic compounds. G. Y. generated transgenic zebrafish line Tg (col2a1-EGFP). N. N. performed other experiments and data analysis. N. N. and W. Z. prepared the manuscript.